ESG data

The Freshfields report suggests that any ESG criteria not harming financial performance should be voluntarily considered. Inevitably, I can only investigate if pension funds’ fiduciary duties prohibit the integration of certain environmental, social or governance criteria. I cannot investigate in a single chapter if pension funds’ fiduciary duties permit the integration of any environmental, social or governance criteria. Hence, I aim for modesty and select a feasible set of environmental, social or governance criteria thereby accepting the inevitable limitation that the investigation of my research question with regard to other ESG criteria will remain a challenge for future research.

Motivated by very large scale corporate environmental disasters (BP’s Gulf of Mexico oil spill, Tepco’s Fukushima nuclear catastrophe), which I expect to concern many pension fund beneficiaries across the world for years to come, I restrict my test to a set of corporate environmental responsibility assessments. Specifically, I employ EIRIS’ assessments in four core processes of corporate environmental responsibility: (i) quality of corporate environmental policy and commitment, (ii) quality of corporate environmental management systems which implement the corporate environmental policy, (iii) improvements of actual environmental performance by corporation as result of the environmental policy and management systems, and (iv) quality of corporate environmental reporting on the previous three processes. All four criteria are assessed by EIRIS on a five point scale. The three quality measurements (environmental policy, environmental management, environmental reporting) are assessed from the worst to the best judgement as ‘inadequate’, ‘weak’, ‘moderate’, ‘good’, or ‘exceptional’ quality of the respective process. The actual environmental performance rating is assessed from the worst to the best judgement as ‘no or inadequate data’, ‘no improvement’, ‘minor improvement’, ‘major improvement’, or ‘significant improvement’. In addition to these four individual (disaggregated) ratings, I calculate the average of these four ratings by transforming the analysis of pension fund ESG integration at the level of the individual stock, it is very likely that they would employ a conceptually very similar research design, since statistical analysis always requires a sufficient high number of individual observations (i.e. ESG integrations at the individual stock level), which can be grouped or

ordinal textual assessments in consecutive integer values following previous studies based on EIRIS data (e.g.Brammer and Pavelin, 2006; Cox et al., 2004; Cox et al., 2007; Dam and Scholtens, 2013). I use this ‘average environmental rating’ as fifth (aggregated) rating, whereby I sort the firms in five groups according to quintiles of the rating scale (i.e. firms rated with values in the smallest 20% of the rating scale are categorised in the worst rated group, companies with values above 20% but no larger than 40% of the rating scale are clustered in the second worst group and so on).

I have access to EIRIS’ end of calendar year assessments from 2003 to 2009 for constituents of the FTSE All World Developed, one of the leading global stock market indexes for developed countries. During my sample period, this index listed companies from 26 developed countries and is hence an ideal investment universe for realistic prudent equity pension fund investment test. These 26 countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Hong Kong, Ireland, Israel (upgraded to developed country in 2008), Italy, Japan, Luxembourg, Netherlands, Norway, New Zealand, Portugal, Singapore, South Korea (upgraded to developed country in 2009), Spain, Sweden, Switzerland, UK, US. This investment universe comprises, on average, around 1,850 firms, whereby a double-digit number of firms are listed with multiple share classes (i.e. A and B shares) each year. EIRIS makes every attempt to provide corporate ESG assessments for each firm in this investment universe, but naturally it needs a bit of operational time to react to each addition to FTSE’s constituent list. This operational time lag effect and some random occasional unavailability of financial data from Datastream resulted in my sample investment universe comprising on average 1,519 firms at the beginning of each year following an EIRIS end of year assessment. The following list shows the annual mean number of firms with available ESG assessments:

2004: 1,504 2005: 1,465 2006: 1,551 2007: 1,520 2008: 1,541 2009: 1,531 2010: 1,519

In columns 7 to 13 in Table 1, I report the actual number of firms with valid ESG data for each portfolio from 2004 to 2010. As EIRIS is specialised in assessing a company's ESG

over time. Whenever new information about a company's ESG performance becomes available EIRIS changes its ratings accordingly. To give one example, when a company has introduced a new environmental policy, then EIRIS will respond to that change by adjusting their rating upwards for that company. Generally, it is common for ESG criteria to change over time. To account for ESG rating changes, I update (re-balance) the constellation of my constructed portfolios based on ESG data at the end of each year. Updating my portfolios annually ensures that companies with improved ESG performance will be upgraded to a portfolio with higher ESG ratings, while companies with worsening ESG performance will be downgraded to a portfolio with lower ESG ratings.